The idea of trusting your business to free software is not one that many FDs take to. The numbers just don’t square. It looks wrong. How can serious, enterprise-level software be free? Where does it come from? Who supports it? Even more importantly, where is it going? Where’s the roadmap? What, in essence, is the risk profile? The answers to these questions have a free-wheeling feel about them that many FDs would find unsettling. The central tenet is that there’s this army of unpaid, geeky enthusiasts who like coding challenges so much that they just want to write the stuff. So don’t worry about a forward development strategy. When the market generates a need, it will be met by this army. That’s how the Open Source model works. So here’s a free operating system for your business … It’s robust. It works. It’s attracting media headlines around the world. So why not give it a whirl? We are talking, of course, about Linux, the Open Source operating system. (“Open Source” means that the source code – the 1s and 0s that allow software to talk to hardware – is “open” for all to see and adjust as they like. This is the opposite of, say, the code for Microsoft Windows NT, which is made available only to trusted partners and only under the strictest of licensing conditions.) Linux was invented back in 1991 by a Finnish computer science graduate, Linus Torvalds, who decided to write a brand new operating system kernel that would use Unix hooks to the outside world. This makes Linux behave like Unix, so it can be readily understood by most of the the army of corporate Unix systems people without much retraining. Developers around the world have taken to Linux like ducks to water because they love the idea of Open Source code that they can play around with and mull over. And Torvalds has opted to license his code under the GNU Public License, which means that any work derived from his Linux kernel – additional programming “modules” or add-ons to Linux – are also Open Source and therefore freely available. The exception to this rule is that applications running on top of Linux are deemed to be independent and not covered by GNU licensing, even if they make calls to (that is, use functionality within) the operating system. With this limitation of the GNU conditions, applications houses can develop software for Linux, since they can still keep control of, and charge for, their products. It is perhaps worth pointing out that Torvalds has also designed the kernel to be able to accept what he calls “kernel modules”, or functional add-ons. So third parties can make additions to the core operating system – like a Windows-style graphical user interface, for example, or clustering, or support for more than one processor in the box. These typically make the operating system either easier to use (for non-technical users who may never have touched Unix-like systems) or more robust for higher-level corporate activities. The Linux community itself decides which modules are worth including and which aren’t through a committee system. Bob Young, CEO of Red Hat, one of the biggest Linux distributors, points out that there are close to 600 Linux components on the Red Hat Linux distribution CD, some 35 of which Red Hat itself puts work into. The logic of having a distributor for a free operating system, he says, comes down to convenience and a fair degree of value-added service. “It is a bit like the car industry. It is perfectly possible for people to go out and gather all the bits that make up a car and assemble these themselves. Most people find it more convenient, however, to buy their car ready-made,” he says. There is a limit to the amount that Red Hat can charge for this service, but when you take into account additional revenues for training and consultancy, there is enough there to constitute a serious revenue stream for Red Hat and three or four other Linux distributors – even if all the components of Linux are free. With investment and support from major IT players such as Intel, Sun, Oracle, Netscape, Computer Associates and Corel, Red Hat is well-placed to provide reassurance to potential customers that “free” doesn’t mean fragile. Dell now offers a Linux installation, which is designed by Red Hat, on new PCs and servers, for example. And all these players are either porting one or more major enterprise-level applications to Linux, or are gifting the operating system with some enterprise level capabilities. As to the lack of a roadmap, Young claims that people who quibble over this put too much faith in the “vapour-ware” pronouncements (promises of functionality which rarely materialise) of mainstream vendors such as Microsoft, and too little faith in the energy and power of the Open Source model. “Open Source means we can rope in the whole user community, all over the world, to debug and stabilise the technology.Today (many research companies) think Linux is far too risky, but hey: six months ago they didn’t even know we existed,” he says. Not everybody agrees with him though. Mark Raphael, a senior analyst with the Meta Group, for example, reckons that his advice to corporate customers is simple: “Linux? Just say ‘No!'” He points out that in Meta Group’s view, while Linux may be just about all right in its present state for technical scientific companies, at the enterprise level it has a long way to go. “There is no substitute for broad support from third party, independent applications vendors. This does not mean just high-profile support from the top-ten vendors in the industry. It means broad-based support from the top 2000 vendors,” he argues. GartnerGroup director of research Simon Hayward, however, takes a slightly more positive view. “Linux requires expertise but there are situations, such as for web servers, or file and print servers, where it is becoming a viable contender for certain enterprise roles. But it has yet to get to the stage where it can be considered a mission-critical enterprise operating system,” he says. It is also the case that, in many companies, systems hacks have already quietly been using Linux for specific tasks, often without the explicit knowledge of senior management. If it works and it’s free, they argue, there’s no reason why it should come to the attention of the “authorities”. And Andy Butler, a senior analyst with the GartnerGroup, says that right now his reservations over Linux have much in common with the kinds of reservations GartnerGroup has about NT. “It lacks the key tools and utilities that data centre managers will demand before they start feeling confident enough to deploy it,” he says. Bob Young views all these analysts’ comments with amusement. “There are three kinds of time in industry today: ordinary time, Net time and Linux time. Let’s see what they’re writing next month,” he says.